Generally, a scanning electron microscope scans an electron beam on a target sample and detects secondary or reflected electrons from the sample to obtain a sample image. If the sample to observe is a thin film of about 100 nm, most of the scanning electron beam transmits the sample. The intensity and scattering angle of the electrons of the beam that transmits the sample depend on the local state, thickness, and atom type of the sample. Such transmission electrons are detected and displayed to obtain the scanning transmission electron microscopy (STEM) image of the sample. The STEM image is divided into bright-field and dark-field images. Different signals are used to detect bright-field images and dark-field images. Bright-field images are obtained by detecting only the electrons that transmit the sample without scattering in the sample. The dark-field images are obtained by detecting the electrons that scatter in and transmit the sample. Generally, a circular detector is used for bright-field images while a torus detector is used for dark-field images.
FIG. 2 shows a block diagram of a general conventional STEM device. The general STEM device is similar in configuration to a transmission electron microscope (TEM) having an acceleration voltage of about 200 kV. A primary electron beam 1 emitted from an electron gun 2 is accelerated by an anode 4, focused by a first focusing lens 5, then scanned on a thin film sample 9 by a deflection coil 8. Electrons that transmit the thin film sample 9 disposed between upper and lower magnetic poles 12a and 12b of an objective lens 11 are focused by a control lens 22 disposed under the objective lens 11 so that the control lens 22 can control their scattering angles detected by a dark-field detector 14. Dark-field signals comprising elastic scattering electrons 13 are thus detected by the dark-field detector 14. Electrons 15 that transmit the dark-field detector 14 are detected by a bright-field detector 17 disposed still under the control lens 22.
The patent document 1 discloses a method for observing a sample using only one detector in which dark-field signals and bright-field signals are switched over by selecting the position of either bright-field image iris or dark-field image iris retained on a common iris base respectively. The patent document 2 discloses a method for enabling a scattering angle range to be selected using a plurality of irises disposed in a plurality of stages. The irises are all similar in mechanism to a camera shutter.    Patent document 1: Official gazette of JP-A No. 169429/1995    Patent document 2: Official gazette of JP-A No. 139988/1994
In recent years, along with the rapid progress of miniaturizing and multi-layering techniques of semiconductors, much attention has been paid to the STEM method that evaluates and measures the miniaturized structures of such semiconductors at their cross sectional views. For normal STEM observations, an acceleration voltage of about 200 kV is used. However, it is found that if an acceleration voltage of 50 kV and under (ex., 30 kV) is used, higher contrast bright-field STEM images are obtained. On the other hand, because the contrast (z contrast) that comes to differ among atomic numbers of the elements of a sample can be observed clearly in dark-field images, the method for observing dark-field images is often combined with sample element analysis (X-ray analysis). Thus, the method is now considered to be very important similarly to the bright-field STEM method.
Usually, the objective lens of a STEM device is designed for observing thin films. As shown in FIG. 3, a thin film sample 9 is disposed approximately in the center between the upper and lower magnetic poles 12a and 12b of the objective lens 11. However, if the STEM device is premised to use the functions of the scanning electron microscope, it is specially structured to observe bulky samples at high resolution. As shown in FIG. 4, therefore, the sample observation plane (place where the primary electron beam is focused) is one-sided to become closer to the upper magnetic pole 12a of the objective lens 11. Consequently, if a STEM observation is to be made for a thin film sample put there, electrons 13 or 15 that transmit the thin film sample come to be affected strongly by the objective lens 11 until they pass the lower magnetic pole 12b of the objective lens 11. The electrons are thus focused again before they pass the lower magnetic pole 12b (FIG. 5). As a result, the electrons transmitting the lower magnetic pole 12b of the objective lens 11 go under the magnetic pole of the objective lens 11 at angles larger than the angles of the electrons scattering in the sample. In addition, if the acceleration voltage of the primary electron beam drops, the angle itself, at which the incident electron beam scatters in the thin film sample, increases. This is why it has been very difficult for conventional techniques to detect electrons scattering at high angles in the target sample, then transmitting the sample if the STEM observation is made using an in-lens type high resolution scanning electron microscope at an acceleration voltage of 50 kV and under. Because, electrons scattering at high angles in the sample, then transmitting the sample come to hit against the inner wall of the magnetic path provided at a halfway before they reach the control lens (C3) 22 disposed under the objective lens 11 or transmission electron detector.